Invited Talk ESA-SRB-ANZBMS 2024 in conjunction with ENSA

Combining AI with medical imaging to diagnose endometriosis earlier and less invasively (#116)

Louise Hull 1
  1. University of Adelaide, Norwood, SA, Australia

Endometriosis negatively impacts 830,000 Australians.  It takes on average 6.4 years and 6 consultations with health professionals before a diagnosis is made. A key barrier to diagnosis is the reliance on surgical visualisation of lesions to obtain a diagnosis. Recent ESHRE and RANZCOG guidelines recognised our Cochrane review findings that diagnostic imaging with specialised transvaginal ultrasound (eTVUS) and MRI (eMRI) scans were the non-invasive diagnostic tests with the most diagnostic potential, recommending their first line use.

IMAGENDO uses artificial intelligence (AI) to combine the diagnostic power of eTVUS and eMRI scans to improve diagnostic accuracy of either eTVUS or eMRI or both. Furthermore, multiple types of lesions can be assessed in our award winning, multimodal, multiple signs AI model.

We intend to transform health care pathways, by using the non-invasive IMAGENDO diagnostic to democratise access to a non-invasive diagnosis of endometriosis in primary care.  We intend to diversify our repository of scans, continuously update our algorithms, develop national and international hubs using federated learning systems.  We also plan to develop automated quality control, credentialing and real time feedback education systems to upskill the workforce. The lack of evidence for primary care treatments of endometriosis can be addressed using IMAGENDO as randomised controlled trials can be conducted for pain management strategies, digital therapeutic approaches, pharmaceuticals and fertility preservation. In the future, general practitioners will have improved evidence based care pathways for endometriosis that provide optimised, targeted preventative strategies to improve quality of life.